Fast decoupled A.C. and A.C./D.C. loadflows.
Thesis DisciplineElectrical Engineering
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
The development of loadflows derived from the basic Newton-Raphson algorithm is reviewed in this thesis with the fast decoupled method emerging as the best in terms of reliability and computation speed. These, and storage requirements, are dependent on programming techniques, and a versatile programme is developed in Chapter 3 using a sparse matrix storage method along with a compatible, semi-optimal dynamically ordered matrix elimination scheme. The same identification vectors are used to locate elements of the admittance and Jacobian matrices of the fast decoupled algorithm. For applications where core computer storage is at a premium, a method is derived, also in Chapter 3, which uses a single Jacobian matrix for both real and reactive power mismatch equations. This gives only slight degradation in reliability. The versatility of the fast decoupled loadflow is extended in Chapter 4 by including the representation of a basic h.v.d.c. interconnection, directly into the Jacobian matrices of the a.c. network. The combined a.c./d.c. Jacobians are solved alternately as in the a.c. method, and it is shown that the reliability, speed and storage advantages offered by the basic fast decoupled loadflow are preserved for large systems. Finally in Chapter 5, a comprehensive d.c. link model is developed which represents all the plant components and operating conditions encountered in practice. The New Zealand a.c.-d.c.-a.c. scheme is used as a test system, and results show convergence for all practical operating conditions without increasing the number of iterations.